Monitoring surface charge migration in the spectral dynamics of single $\mathrm{Cd}\mathrm{Se}∕\mathrm{Cd}\mathrm{S}$ nanodot/nanorod heterostructures

Spherical CdSe nanocrystals capped by a CdS rod-like shell exhibit interesting spectral dynamics on the single particle level. Spectral boxcar averaging reveals a high degree of correlation between the emission energy, spectral linewidth, phonon coupling strength, and emission intensity of the single nanocrystal. The results can be described in terms of a spatially varying surface charge density in the vicinity of the exciton localized in the CdSe core, leading to a quantum confined Stark effect which modifies the transition energy and the radiative rate. Whereas internal charging of the particle results in a change in the nonradiative rate, surface charges primarily influence the radiative rate. Additionally, we observe characteristic spectral dynamics in frequency space, the magnitude of which depends slightly on temperature and strongly on excitation density. By distinguishing between continuous spectral jitter and discrete spectral jumping associated with a reversible particle ionization event, we can attribute the spectral dynamics to either a slowly varying surface charge density or a rapidly occurring polarization change due to a reversible expulsion of a charge carrier from the semiconductor nanostructure. Whereas the former exhibits universal Gaussian statistics, the latter is best characterized by a Lorentzian noise spectrum. The Gaussian spectral noise increases with spectral redshift of the emission and with increasing proximity of the surface charge to the localized exciton. The observation of a high degree of correlation between peak position and linewidth right up to room temperature suggests applications of the nanocrystals as extremely sensitive single charge detectors in both solid state devices and in biomolecular labeling, where highly local measurements of the dielectric environment are required. Nanoscale control of the physical shape of nanocrystals provides a versatile test bed for studying electronic noise, making the approach relevant to a wide range of conducting and emissive solid state systems.

Figure 1

(Color online) (a)–(d) Transmission electron microscopy images of 4 nm CdSe NCs [(a),(c)] and elongated CdSe/CdS nanohererostructures [(b),(d)] formed by epitaxial growth of CdS nonorods on the $\left\{00\overline{1}\right\}$ facets of CdSe nanocrystals. The location of the emitting CdSe core inside the CdSe/CdS nanoheterostructure is indicated by a dashed circle in (b). (e) Spectral diffusion in the fluorescence of single elongated CdSe/CdS nanocrystals at $5\mathrm{K}$, $50\mathrm{K}$, and $300\mathrm{K}$. The PL spectra are plotted as a function of time in a two-dimensional representation with the intensity encoded in color/gray scale (blue/very dark gray: low intensity, red/dark gray: high intensity). The temporal resolution of the data is $1\mathrm{s}$.

Figure 2

(Color online) Correlation between peak position and peak linewidth for three different particles recorded at $5\mathrm{K}$, $50\mathrm{K}$, and $300\mathrm{K}$. (a) Two representative spectra taken from the traces in Fig. 1, illustrating a spectral broadening concomitant with the spectral redshift. (b) Correlation between spectral linewidth and peak position. The open squares correspond to the raw data. The solid squares represent filtered data sets, in which only spectral shifts were considered which occur on time scales longer than the measurement resolution of 1 s. The large solid squares indicate a binned average.

Figure 3

(Color online) PL spectra of one single NC recorded during the spectral diffusion process and sorted by peak energy. As the peak energy shifts to the red, the LO phonon sideband increases in intensity and the spectra broaden, giving rise to a funnel-like shape of the graph (a). (b) The effect is also visible at room temperature, although the phonon sideband can no longer be resolved. A redshift in the emission results in an unambiguous spectral broadening.

Figure 4

Correlation between fluorescence peak position and linewidth (FWHM), phonon coupling strength, and PL intensity for the sorted spectral trace measured at $5\mathrm{K}$ and shown in Fig. 3a. Sorting of the spectra allows us to bin related spectra prior to fitting and extracting the linewidth, peak position, phonon coupling, and intensity. The scheme illustrates the mechanism responsible for the correlations, explained in more detail in Fig. 10. Briefly, the hole is confined in the CdSe core, whereas the electron is free to penetrate the CdS shell. Surface charges modify the overlap of the electron and hole wave functions, which in turn control FWHM, phonon coupling, peak energy, and PL intensity.

Figure 5

(Color online) Spectral diffusion statistics of the PL trace shown in Fig. 1a. The histograms illustrate the difference in energy $\Delta E$ between two subsequent spectral peaks. (a) Overall trace consisting of 800 events. (b) Events interrupted by fluorescence intermittency (total of 90). The insets show the histograms on a logarithmic scale with a Gaussian (solid line) and a Lorentzian (dotted line) superimposed.

Figure 6

Temperature dependence of the spectral diffusion statistics. (a) Cumulative $\Delta E$ distribution extracted during continuous emission of 15 $\left(5\mathrm{K}\right)$, 17 $\left(50\mathrm{K}\right)$, and 16 $\left(300\mathrm{K}\right)$ single particles. (b) $\Delta E$ distribution extracted from energetic jumps following a blinking event.

Figure 7

Intensity dependence of spectral diffusion at $5\mathrm{K}$. (a) Cumulative $\Delta E$ distribution extracted from 18 single particles during continuous emission. (b) $\Delta E$ distribution extracted from energetic jumps following a blinking event.

Figure 8

Dependence of the width of the $\Delta E$ histogram in both continuous spectral diffusion (circles) and discrete spectral switching (squares) for an average of 18 single particles at $5\mathrm{K}$.

Figure 9

(Color online) Dissection of the spectral diffusion of a single NC at $5\mathrm{K}$ into three regions, depending on the magnitude of the Stark effect. (a) The linewidth-peak energy correlation appears continuous and is arbitrarily divided into three equally large regions. (b) $\Delta E$ histograms of the continuous spectral diffusion (blinking events discarded) of the three regions indicated in panel (a). (c) $\Delta E$ distribution of Sec. 3 shown on a logarithmic scale with a Gaussian superimposed.

Figure 10

Schematic representation of the influence of surface charge density on the emission from a single elongated NC. (a) Surface charge present at two different distances from the hole trapping core of the NC results in different magnitudes of a Stark shift in the emission. As the detection of the emission spectrum occurs over a finite time, spatial jitter in the charge density will result in spectral broadening. The closer the surface charge is located to the core of the NC, the stronger the redshift and the stronger the line broadening. (b) The electric field $F$ induced by the surface charge also results in a perturbation of the electron-hole wave function overlap within the NC and thus of the radiative rate. A large local electric field leads to a displacement of the electron with respect to the hole and, therefore, reduces the wave function overlap and the radiative rate, leading to a reduction in PL intensity under the premise of constant nonradiative decay.